/// <summary>
/// Collects all of the data that needs to be indexed as defined in the index set.
/// </summary>
/// <param name="node">Media item XML being indexed</param>
/// <param name="type">Type of index (should only ever be media)</param>
/// <returns>Fields containing the data for the index</returns>
protected override Dictionary <string, string> GetDataToIndex(XElement node, string type)
{
var fields = base.GetDataToIndex(node, type);
//find the field which contains the file
XElement fileElement = node.Elements().FirstOrDefault(x =>
{
if (x.Attribute("alias") != null)
{
return((string)x.Attribute("alias") == this.UmbracoFileProperty);
}
else
{
return(x.Name == this.UmbracoFileProperty);
}
});
//make sure the file exists
if (fileElement != default(XElement) && !string.IsNullOrEmpty(fileElement.Value))
{
// Parse the current value
string filePath = fileElement.Value;
if ((filePath).StartsWith("{"))
{
filePath = JObject.Parse(filePath).Value <string>("src");
}
if (!filePath.IsNullOrWhiteSpace())
{
// Get the file path from the data service
string fullPath = this.DataService.MapPath(filePath);
var fi = new FileInfo(fullPath);
if (fi.Exists)
{
try
{
if (!SupportedExtensions.Select(x => x.ToUpper()).Contains(fi.Extension.ToUpper()))
{
DataService.LogService.AddInfoLog((int)node.Attribute("id"), "UmbracoExamine.FileIndexer: Extension '" + fi.Extension + "' is not supported at this time");
}
else
{
fields.Add(TextContentFieldName, ExtractTextFromFile(fi));
}
}
catch (Exception ex)
{
DataService.LogService.AddErrorLog((int)node.Attribute("id"), "An error occurred: " + ex);
}
}
else
{
DataService.LogService.AddInfoLog((int)node.Attribute("id"), "UmbracoExamine.FileIndexer: No file found at path " + filePath);
}
}
}
return(fields);
}
private void buttonBrowse_Click(object sender, EventArgs e)
{
var wildcards = SupportedExtensions.Select(p => "*" + p);
var dialog = new OpenFileDialog()
{
Filter = FileFilterStrings.AudioFilter + string.Format("({0})|{1}", string.Join(", ", wildcards), string.Join(";", wildcards))
};
if (dialog.ShowDialog(this) == DialogResult.OK)
{
filePathBox.Text = dialog.FileName;
}
}
/// <summary>
/// need to prevent extensions not in allowed extensions getting into the index
/// </summary>
/// <param name="node"></param>
/// <returns></returns>
private bool IsAllowedExtension(XElement node)
{
var extension = node.Element("umbracoExtension");
if (extension != null)
{
var extensionValue = "." + extension.Value;
if (SupportedExtensions.Select(x => x.ToUpper()).Contains(extensionValue.ToUpper()))
{
return(true);
}
}
return(false);
}
/// <summary>
/// Provides the means to extract the text to be indexed from the file specified
/// </summary>
/// <param name="file"></param>
/// <returns></returns>
protected virtual string ExtractTextFromFile(FileInfo file)
{
if (!SupportedExtensions.Select(x => x.ToUpper()).Contains(file.Extension.ToUpper()))
{
throw new NotSupportedException("The file with the extension specified is not supported");
}
var mediaParser = new MediaParser();
Action <Exception> onError = (e) => OnIndexingError(new IndexingErrorEventArgs("Could not read media item", -1, e));
var txt = mediaParser.ParseMediaText(file.FullName, onError, out _extractedMetaFromTika);
return(txt);
}
/// <summary>
/// Provides the means to extract the text to be indexed from the file specified
/// </summary>
/// <param name="file"></param>
/// <returns></returns>
protected virtual string ExtractTextFromFile(FileInfo file)
{
if (!SupportedExtensions.Select(x => x.ToUpper()).Contains(file.Extension.ToUpper()))
{
throw new NotSupportedException("The file with the extension specified is not supported");
}
var pdf = new PDFParserPdfBox();
Action <Exception> onError = (e) => OnIndexingError(new IndexingErrorEventArgs("Could not read PDF", -1, e));
var txt = pdf.GetTextFromAllPages(file.FullName, onError);
return(txt);
}
/// <summary>
/// Provides the means to extract the text to be indexed from the file specified
/// </summary>
/// <param name="filePath"></param>
/// <param name="mediaFileSystem"></param>
/// <returns></returns>
protected virtual string ExtractTextFromFile(string filePath, MediaFileSystem mediaFileSystem)
{
var fileExtension = mediaFileSystem.GetExtension(filePath);
if (!SupportedExtensions.Select(x => x.ToUpper()).Contains(fileExtension.ToUpper()))
{
throw new NotSupportedException("The file with the extension specified is not supported");
}
var pdf = new PDFParser();
Action <Exception> onError = (e) => OnIndexingError(new IndexingErrorEventArgs("Could not read PDF", -1, e));
var txt = pdf.GetTextFromAllPages(filePath, mediaFileSystem, onError);
return(txt);
}
public SoundSourceSelector()
{
InitializeComponent();
AllowDrop = true;
browseButton.Click += (s, e) =>
{
var wildcards = SupportedExtensions.Select(p => "*" + p);
var dialog = new OpenFileDialog()
{
Title = "音源選択",
Filter = "音声ファイル" + string.Format("({0})|{1}", string.Join(", ", wildcards), string.Join(";", wildcards))
};
if (dialog.ShowDialog(this) == DialogResult.OK)
{
fileBox.Text = dialog.FileName;
}
};
}